https://bugs.documentfoundation.org/show_bug.cgi?id=142635

Mike Kaganski <[email protected]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[email protected]

--- Comment #16 from Mike Kaganski <[email protected]> ---
The problem happens in lcl_ScAddress_Parse_OOo
(sc/source/core/tool/address.cxx) during *copy* to clipboard. The 'if (!bExtDoc
&& !rDoc.GetTable( aTab, nTab ))' branch executes (because rDoc doesn't happen
to have the table (sheet) with the "Sheet.1" name), and then for the testcase
from comment 0, the 'if (n > 0 && aTab.getLength() - n <= 4)' executes, as if
".1" was a file extension, making the import code think that it was an external
reference (for names without a dot, or with more than 4 characters after the
dot, it would be different - likely manifesting as comment 15 describes).

The copy code has at least three ScDocument live on the call stack: the
original document, the clipboard document, and some temporary document used in
Chart-specific code here; the latter has no correct table name (yet?).

I suppose that the range a2e6f31c6f90e446d1462e7c80f6b1317f7825bc ..
b9652803096b68a33702601aac52e78c8a4250c6 could change something that the
platform-specific clipboard code now could do something with the Chart-specific
ScDocument so that it already has correct table names when this code executes,
while Windows-specific code does that in a different order. However, the
multiple "works here on Windows" responses look confusing, given how
consistently reproduce the problem on Windows on my system (and the test
failing for me locally - that I disabled in the commit mentioned in comment 2 -
at least worked fine on Jenkins Windows builders).

Eike: do you have an idea where could I look at to find the problem?

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to